[606] in Commercialization & Privatization of the Internet
Some Thoughts About NREN Telecom Policy
daemon@ATHENA.MIT.EDU (tmn!cook@uunet.uu.net)
Wed Apr 17 02:30:09 1991
From: tmn!cook@uunet.uu.net
To: com-priv@psi.com
Date: Wed Apr 17 00:54:02 1991
<<MESSAGE from>> Gordon Cook 17-APR-91 0:54
cook@tmn
Could I try to rachet some of this discussion into some very broad
generalizations of the kind that please policy analysts?
What if NREN were viewed as nothing more than a broadband digital
distribution channel, as happy to deliver voice mail as it is to deliever
distributed supercomputing, and spreadsheets as it is to deliver knowbots?
If you take it down a notch from your favorite application to its lowest
common denominator does it help in thinking about whether it is easy to
place boundaries around it? (As in it's just a network for scientists?)
And in whether you can expect the hardware and software produced to
support it to stay within the narrow confines of a science network??
And what about deployment scenarios? How many are there? Can an argument
be made that they can all be shoved into two possibilities? Option A
being a private, wide area network, possibly using proprietary protocols
running on top of SONET? Built on dumb bit pipes from the telcos but
other than this duplicating whatever infrastructure the telcos may begun
to place in the field in order to spread offerings of SMDS and which they
will eventually change into B-ISDN?
Option B being a network based on telco oriented protocols such as ATM
over SONET and SMDS and B-ISDN. Here NREN switches would be developed
presumably either by the telcos or by companies like Proteon, BBN,
Wellfleet, cisco, etc using CCITT B-ISDN standards at the first 3 levels
of the stack and TCP or whatever replaces it at level 4? The gov'ts
investment here presumably comes in synergy with the Public Switched
Telephone Network's efforts rather than in duplication of them?
Are there other broad brush options that escape me? What do you think
will decide, or should decide which one is used to implement NREN?
How is one to judge whether problems with internetting a gigabit speeds
might mandate the selection of one variety of switch for the entire
network? Is it correct that the Japanese and Europeans are producing
functional multi-hundred megabit per second switches based on ATM over
SONET? IE the telco scenario? If so, is there any danger that from the
point of view of international economic competitiveness we could build a
debilitating wall around ourselves if we didn't go with the telco scenario
of deployment?
Or should we assume that switches that support international real time
cad/cam applications and real time video will be developed quite
independently of the NREN effort and that, for example, an effort to
support the design of Boeing's 777 aircraft will have nothing to do with
NREN technology?
Thanks, Gordon Cook - Office of Technology Assessment
